The Tridge ontology

HS codes were built for tariffs. We built a classification for markets.

Mango, guava and mangosteen share HS 0804.50. They do not share a season, a buyer, a price or a risk. Tridge maintains its own agri-food ontology so each of them can be analysed as the market it actually is.

The problem with tariff codes

A code designed to collect duty cannot describe a market.

The Harmonized System exists so customs authorities can assess tariffs consistently. That is a different job from telling a buyer what a market did. Where the duty is the same, the code merges — and agri-food is where it merges most.

Tariff-purpose classification

080450

Guavas, mangoes and mangosteens, fresh or dried

Three fruits. One line in the trade statistics.

Tridge ontology — three markets

Fresh Mango/market-overview/fresh-mangoFresh Guava/market-overview/fresh-guavaFresh Mangosteen/market-overview/fresh-mangosteen

Look up 0804.50 in any tariff schedule, then open the three pages. The claim is checkable.

What an ontology is

A classification, plus the relationships between its terms.

A list of names is a taxonomy. An ontology also records how the names relate — which product is a variety of which, which is processed from which, and which tariff line each one is filed under. Those relations are what let a question about one product pull in evidence about another.

ITEM

Fresh Mango

The market a buyer actually trades — one product, one season, one price series.

SUB-PRODUCTS

10 varieties

Alphonso, Amrapali, Ataulfo, Dasheri, Haden, Keitt, Kent, Kesar, Palmer, Tommy Atkins.

DERIVED

7 processed goods

Dried mango, frozen mango, mango puree, mango juice concentrate, pickapeppa sauce and more.

HS MAPPING

080450

The tariff line it is filed under — kept as a mapping, not as the identity.

Every one of those relations is already published on the Fresh Mango market page. The ontology is not a claim about an internal system; it is what those pages are made of.

Open the Fresh Mango market page →

What we classify

Two things, not one — products and companies.

PRODUCTS

Item-level, agri-food-native

8,400 products with data, each carrying its varieties, its processed derivatives, its units and grades, and the tariff lines it maps to.

  • Varieties separated where the market prices them separately
  • Processed goods linked back to their raw input
  • Units, currencies and grades normalised so series compare

COMPANIES

Identity resolved across customs sources

The same exporter is filed under different spellings, abbreviations and legal suffixes in every system that records it. 730,000+ company profiles carry one resolved identity and the trade history attached to it.

  • Spellings, aliases and suffixes collapsed into one entity
  • Trade footprint, partners and product mix per company
  • Companies indexed against the same product ontology

What it unlocks

Six datasets, one classification, one question.

Size is not the argument — plenty of vendors have volume. The argument is that all of it lands on the same classification, so a question about one product can be answered with evidence from all six at once.

35M+

Wholesale & farmgate prices

Domestic price signals at origin, in normalised units and currencies.

220M+

Import & export transactions

Trade records as of Aug 2026: transaction-level international trade since 2018, with counterparties in 230 countries and territories.

HS

Tariffs

Duty lines mapped to the ontology, so landed cost follows the product.

NDVI

Weather

Conditions over the growing regions that actually supply the product.

Daily

News & reports

Speaker reports tagged to the same products and countries.

730,000+

Company identity

Who shipped what, resolved across the systems that filed it.

See it in the open

Check it against your own product.

The ontology is not a diagram in a deck. It is the structure behind 8,400 products with data — each one published with its sub-products, its derived products and the HS code it maps to. Pick a product and see for yourself.

How we maintain it

A classification is only as good as its upkeep.

01

Sourcing

Data comes from a network of global suppliers and official sources, each reviewed by product and market specialists.

02

Validation

Statistical checks flag outliers and inconsistencies before anything enters the analysis layer.

03

Mapping

Records are mapped onto the ontology — product, variety, derivative, company identity — with specialist review where the model is uncertain.

04

Monitoring

Source quality is scored continuously; a source that drifts is dropped rather than averaged in.

Where it shows up

Everything downstream inherits it.

SHELF

Comparable series

Prices and flows line up because the product underneath is the same product.

COMPANY ANALYTICS

Ledgers you can read

A resolved company can be aggregated, ranked and compared — and its shipments attributed to the right product.

AI CURATION

Answers with reasoning

A board can cross-check sections because it knows how the products relate.

TRIDGE ARK

Targeting that fits

Counterpart matching runs on category fit, not on a keyword in a company name.

Questions

The classification, answered.

Why do mango, guava and mangosteen share the same HS code?

HS 0804.50 groups them because they attract the same tariff treatment, not because they are the same market. The Harmonized System is a duty schedule; grouping by duty is exactly what it is for. It just means trade statistics for that code describe none of the three on its own.

Can you analyse agricultural trade data using HS codes alone?

Only coarsely. Wherever a code merges distinct products — and in agri-food it merges constantly — an HS-level series mixes different seasons, buyers and price levels into one number. You can see totals move; you cannot attribute the move.

What is an agri-food product ontology?

A classification of food and agricultural products plus the relationships between its terms: which product is a variety of which, which is processed from which, and which tariff line each maps to. Tridge maintains one and uses it to normalise prices, trade, tariffs, weather and news.

How does Tridge classify products differently from HS codes?

At item level, and with relations. Fresh Mango is its own market, its ten named varieties sit beneath it, its seven processed derivatives link back to it, and HS 0804.50 is recorded as a mapping rather than as the product’s identity.

How does Tridge match the same company across different customs systems?

Through identity resolution: spellings, abbreviations, legal suffixes and transliterations that refer to one business are collapsed into a single entity, which the trade history then attaches to. That is what makes 730,000+ company profiles comparable instead of being 730,000+ name strings.

The fastest way to judge a classification is to use it.

Open a product market and check the sub-products, the derived products and the HS mapping against what you know about your own category.

By clicking “Accept Cookies,” I agree to provide cookies for statistical and personalized preference purposes. To learn more about our cookies, please read our Privacy Policy.